Van der kolk.pm6

A Randomized Clinical Trial of
Eye Movement Desensitization and
Reprocessing (EMDR), Fluoxetine, and Pill Placebo
in the Treatment of Posttraumatic Stress Disorder:
Treatment Effects and Long-Term Maintenance
Bessel A. van der Kolk, M.D.; Joseph Spinazzola, Ph.D.;
Margaret E. Blaustein, Ph.D.; James W. Hopper, Ph.D.; Elizabeth K. Hopper, Ph.D.;
Deborah L. Korn, Psy.D.; and William B. Simpson, Ph.D.
Received May 27, 2005; accepted May 25, 2006. From the Trauma Center, Justice Resource Institute, Brookline (all authors); the BostonUniversity School of Medicine, Boston (Drs. van der Kolk, Spinazzola, Objective: The relative short-term efficacy and
Blaustein, and E. Hopper); McLean Hospital, Belmont (Dr. J. Hopper); long-term benefits of pharmacologic versus psycho- and Boston Research Associates, Boston (Dr. Simpson), Mass. therapeutic interventions have not been studied for Supported by grant R01MH58363 from the National Institute posttraumatic stress disorder (PTSD). This study compared the efficacy of a selective serotonin reup- Presented at the 19th annual meeting of the International Society for Traumatic Stress Studies, October 29–November 1, 2003, Chicago, take inhibitor (SSRI), fluoxetine, with a psychothera- Ill., and at the 157th annual meeting of the American Psychiatric peutic treatment, eye movement desensitization and Association, May 1–6, 2004, New York, N.Y. reprocessing (EMDR), and pill placebo and measured Dr. Korn has served on the speakers or advisory boards for and maintenance of treatment gains at 6-month follow-up.
received honoraria from the EMDR International Association and Method: Eighty-eight PTSD subjects diagnosed
the EMDR Institute, Inc. Drs. van der Kolk, Spinazzola, Blaustein, according to DSM-IV criteria were randomly as- J. Hopper, E. Hopper, and Simpson report no additional financialaffiliations or other relationships relevant to the subject of this article. signed to EMDR, fluoxetine, or pill placebo. They Acknowledgments appear at the end of this article. received 8 weeks of treatment and were assessed by Corresponding author and reprints: Bessel van der Kolk, M.D., blind raters posttreatment and at 6-month follow-up.
Trauma Center, 1269 Beacon St., Brookline, MA 02446 The primary outcome measure was the Clinician- Administered PTSD Scale, DSM-IV version, andthe secondary outcome measure was the Beck De-pression Inventory-II. The study ran from July 2000 xposure to traumatic experiences is ubiquitous in Eour society: 60% of men and 51% of women in the
Results: The psychotherapy intervention was
general population report at least 1 traumatic event in more successful than pharmacotherapy in achievingsustained reductions in PTSD and depression symp- their lives.1 For men, combat and witnessing injury or toms, but this benefit accrued primarily for adult- death are the most frequent precipitants for developing onset trauma survivors. At 6-month follow-up, 75.0% posttraumatic stress disorder (PTSD), while for adult of adult-onset versus 33.3% of child-onset trauma women, physical attacks by intimate partners is the most subjects receiving EMDR achieved asymptomatic frequent cause. Approximately 9.8 million adult Ameri- end-state functioning compared with none in thefluoxetine group. For most childhood-onset trauma can women (10.3%) have histories of violent physical patients, neither treatment produced complete symp- assaults, and 12.1 million (12.7%) have experienced a completed rape at some point in their lives.2 More than Conclusions: This study supports the efficacy
twice as many women report histories of childhood sexual of brief EMDR treatment to produce substantial abuse than of adult rape,3 which occurs in approximately and sustained reduction of PTSD and depressionin most victims of adult-onset trauma. It suggests a 10% of the general population.1 Childhood sexual abuse role for SSRIs as a reliable first-line intervention to is a strong predictor of subsequent PTSD.4,5 More than achieve moderate symptom relief for adult victims of 20% of returning veterans from Iraq are currently seeking childhood-onset trauma. Future research should as- sess the impact of lengthier intervention, combination A variety of psychotherapeutic approaches to PTSD, treatments, and treatment sequencing on the resolu- tion of PTSD in adults with childhood-onset trauma.
all involving some form of exposure and “trauma process- (J Clin Psychiatry 2007;68:00–00) ing,” have been shown to be effective (e.g., see references7–10). Pharmacologic agents, in particular the selectiveserotonin reuptake inhibitors (SSRIs), have also been shown to be effective in civilians with PTSD, but less so in effective evidence-based treatment for PTSD in the profes- veterans (e.g., see references 11–15). However, no study sional treatment guidelines of the U.S. Department of De- has directly compared the efficacy of psychotherapeutic fense,28 the American Psychological Association,29 and the The relative efficacy of biological versus psychothera- Comparing the relative effectiveness of pharmacother- peutic interventions is an issue of great clinical and eco- apy versus exposure therapy is particularly relevant in the nomic relevance. Comparative studies have been done in treatment of PTSD, since the SSRIs are widely used to panic disorder,16–18 obsessive-compulsive disorder,19,20 and treat PTSD, particularly in primary care and health mainte- depression.21,22 Most studies in those disorders conclude nance organization (HMO) settings, where little attention that cognitive-behavioral therapy (CBT) and pharmaco- may be paid to helping patients “process” their traumas.
therapy are roughly equivalent, but there continues to be This is reasonable as long as the question has not been considerable controversy about the interpretation of the settled whether pharmacotherapy produces better or worse results than the active processing of traumatic memories A recent meta-analysis of a large depression sample25 (which many patients are reluctant to do because of want- found that “among those with a history of early childhood ing to avoid being reminded of their trauma).31 Our study trauma (physical or sexual abuse, neglect, or loss of par- was particularly concerned with the effects of these vari- ents at an early age), psychotherapy alone was superior ous treatments on clinical symptomatology over time.
to antidepressant monotherapy.”(p14293) Treatment with psy-chotherapy resulted in twice the rate of remission in patients with major depression who also had histories of serious early adverse life events than did treatment with antidepressants. Surprisingly, the prevalence of PTSD was Following institutional review board approval, individ- not ascertained in that sample. Given the high rate of uals 18 to 65 years old, with current PTSD and with mixed PTSD comorbidity with depression,1–4 the PTSD outcome trauma exposure at least 1 year prior to intake, were re- literature may make an important contribution to effective cruited via newspaper ads, the Internet, and solicitation from medical and mental health professionals. The study Research suggests that PTSD patients are more respon- ran from July 2000 through July 2003. A total of 229 par- sive to treatments that specifically “process” traumatic ticipants were assessed at pretreatment after giving written memories than to either supportive counseling or stress informed consent. Of these, 88 (38%) met study inclusion inoculation training.8 In 1 meta-analysis of 61 treatment criteria and were randomly assigned to treatment. Of the outcome trials for PTSD (including drug therapies, CBT, remainder, 47 (21%) failed to meet DSM-IV diagnostic eye movement desensitization and reprocessing [EMDR], criteria for PTSD, 30 (13%) withdrew consent prior to relaxation training, hypnotherapy, and dynamic therapy), randomization, and 64 (28%) met study exclusionary crite- psychological therapies were more effective than drug ria. Baseline participant information is contained in Table therapies, and both were more effective than controls.26 1. Trauma history was obtained by self-report. The pre- Psychological therapies had lower dropout rates than phar- dominant index trauma involved interpersonal victimiza- macotherapies (14% vs. 32%). Among the psychological tion (71.6%); in 50% of participants, trauma onset oc- approaches, CBT and EMDR were the most effective, with no demonstrable differences in treatment efficacy.
Exclusion criteria were unstable medical condition, Among the drug therapies, the SSRIs had the greatest ef- contraindications to either treatment (i.e., pregnancy, glau- fect sizes. Another effect-size analysis of 14 treatment out- coma or detached retina, or history of severe allergies or come studies27 found that the drug therapies with the larg- multiple adverse drug reactions), inability to be weaned est effect sizes were fluoxetine and amitriptyline.
off current psychotropic medications, psychotic or bipolar The present study directly compared the efficacy of a disorder, current alcohol or substance abuse/dependence, psychopharmacologic agent, fluoxetine, with a psycho- severe dissociation, active suicidality or life-threatening therapeutic, exposure-based treatment, EMDR, and a pill mutilation, prior exposure to active study interventions, placebo in PTSD. EMDR is an exposure treatment in concurrent trauma-focused treatment, unstable living situ- which patients perform saccadic eye movements while ation, Global Assessment of Functioning32 score < 40, and thinking about a traumatic experience. Rather than pro- disability compensation for PTSD or pending trauma- viding a chronological narrative of the details of the trau- related lawsuit. Initial telephone screening was used to as- matic event, as is done in CBT, EMDR patients are sess likely presence/absence of inclusion and exclusion encouraged to follow their own course, moving freely criteria; potential participants were then invited for in- backward and forward in time, attending to inner sen- person assessment. Participants were not excluded for sations and cognitions, omitting verbal communication engagement in nontrauma-focused, supportive psycho- about content if they wish. EMDR has been declared an therapy, provided this treatment had been ongoing for Table 1. Baseline Analyses of Demographic Variables and Comorbid Diagnoses by Treatment Group
No. of current comorbid Axis I/II diagnoses, mean (SD)c aTotal of all 3 treatments combined.
bUsed omnibus analysis of variance for continuous measures or Pearson χ2 statistic for categorical measures.
cAssessed using Structured Clinical Interview for DSM-IV Disorders.37,38Abbreviations: BDI-II = Beck Depression Inventory-II, CAPS = Clinician-Administered PTSD Scale, EMDR = eye movement desensitization and reprocessing, PTSD = posttraumatic stress disorder.
Beck Depression Inventory-II (BDI-II).39 This widely
Table 2. Index Trauma Distribution (N = 88)
used and psychometrically sound self-report measure of depressive symptoms was used to assess secondary study hypotheses regarding the immediate and long-term effi- cacy of pharmacologic versus psychological interventions on amelioration of depressive symptoms in traumatized Assessment.
ment, posttreatment, and 6 months following treatment cessation. Assessment included interview and self-report at least 3 months prior to study baseline and did not in- measures, as well as psychophysiologic response to script- volve exposure to or processing of traumatic memories. A driven imagery. This was a modified version of the proto- detailed report of screening attrition and exclusion data is col adapted by Pitman and colleagues40 for use with Review available.33
PTSD, consisting of a series of 30-second, audio-tapedscripts in second-person, present-tense narrative, based on participants’ recounting of the worst memory associated Clinician-Administered PTSD Scale (CAPS), DSM-IV
with the trauma, as well as a neutral memory from the Version.34 The CAPS total score was the primary
same time period. Modifications involved use of 2 alter- continuous outcome measure as an index of PTSD symp- nating sets of neutral and traumatic scripts. Physiologic Only For tom severity. Rating was based on a 1-week interval for
responses to script-driven imagery before and after treat- immediate pretreatment and posttreatment assessment; a 1-month interval was used at baseline and 6-month Evaluators were primarily postdoctoral-level clinicians follow-up. Results are also presented on PTSD diagnostic who received extensive training and ongoing supervision Review status, defined as full DSM-IV diagnostic criteria and
in administration of study measures. Interrater reliability using the following CAPS scoring rules: (1) total severity on the CAPS was established prior to the start of the study, score > 50, (2) per-item frequency of at least 1 and inten- based on coding of live and videotaped interviews, and re- sity of at least 2, and (3) per-item total severity score of assessed at regular intervals to avoid rater drift. Interrater at least 4.35 Asymptomatic end-state function (i.e., PTSD reliability for PTSD diagnosis, based on Cohen’s kappa, symptom remission), defined as CAPS score < 20, is also was very good (κ = 0.82, percent agreement = 0.92).
Interrater reliability for PTSD symptom severity was Structured Clinical Interview for DSM-IV Axis I and
excellent (intraclass correlation coefficient = 0.96). All Axis II Disorders (SCID I37 and SCID II38). This struc-
raters were blind to treatment condition and were never tured interview was used for determination of PTSD and assigned the same participant for both pretreatment and Treatment. Participants were randomly assigned to 1
discontinued over a 10-day period following the 8-week of 3 treatment conditions: EMDR, fluoxetine, or pill pla- intervention period. Pharmacotherapists were licensed cebo. Randomization was stratified by presence/absence psychiatrists (fourth-year residents and above) who re- of concurrent supportive psychotherapy. In order to ensure ceived extensive training on the study protocol and re- approximately equal numbers in each treatment condition, ceived weekly supervision by the principal investigator.
random assignment was blocked in groups of 12 consecu-tive participants, so that in each block, 4 participants were assigned to each condition. Participants in all 3 conditions The study was designed to evaluate whether both received a total of 8 weekly treatment sessions. Following fluoxetine and EMDR perform equally well, and better cessation of treatment, participants in the 2 active treat- than placebo, after 8 weeks of treatment of PTSD and ment conditions were asked to refrain from initiating new whether EMDR differs from fluoxetine in maintaining treatment during the 6-month follow-up period. The blind treatment gains over time. Secondary hypotheses ad- was removed from the placebo intervention following the dressed the impact of child- versus adult-onset trauma on posttreatment assessment, and, for ethical reasons, partici- treatment outcome. The impact of treatment on depressive pants were offered the option of receiving either of the 2 active treatments. Consequently, placebo group data were Study hypotheses were tested on both treatment com- pleter and intent-to-treat (ITT) samples. Intent-to-treat EMDR. EMDR treatment consisted of 90-minute indi- analyses were conducted using an early termination as- vidual sessions. The treatment targeted memories associ- sessment, when available, or a last-observation-carried- ated with the primary trauma identified during pretreat- forward (LOCF) procedure to impute missing data. Long- ment evaluation. Treatment was administered according term effects of active-treatment completers were assessed to a manual developed for this study,41 based on the stan- at 6 months posttreatment for both follow-up completer dard protocol.42 The 4 clinicians in the EMDR condition and intent-to-follow (ITF) samples. For the ITF sample, were at master’s level or above, with Level II EMDR missing follow-up data were also estimated using LOCF.
training. They were licensed in their profession and had Baseline group differences were assessed using anal- a minimum of 3 years’ experience treating patients with ysis of variance for continuous measures and Pearson χ2 PTSD. Clinicians received extensive training and bi- statistic for categorical measures. To minimize bias, post- weekly supervision in the manualized protocol established treatment continuous-measure analyses were conducted for the study from a certified senior EMDR instructor using omnibus analysis of covariance (ANCOVA) with (D.L.K.). An average of 6 EMDR sessions were devoted baseline as the covariate.43,44 These were followed by pair- to trauma processing. All EMDR sessions were video- wise comparisons using 2-group ANCOVA. Baseline by taped, and an independent evaluator assessed treatment treatment-condition interactions were tested for all pri- fidelity through videotape review from randomly sampled mary outcome analyses. No significant interactions were found. Therefore, interaction terms were dropped from Medication. Medication treatment consisted of 20- to further analyses. Follow-up analyses were conducted on 30-minute individual sessions. The fluoxetine and placebo the 2 active treatments. Analyses of continuous variables interventions were administered in a double-blind, fixed- used ANCOVA, controlling for baseline score. Posttreat- flexible-dose design according to standard protocol for ment and follow-up categorical outcomes (i.e., PTSD double-blind pharmacologic interventions for PTSD.12–15 diagnosis and end-state function) were evaluated using This study employed a manualized pharmacotherapy pro- the Pearson χ2 statistic. Within-group and between-group tocol developed and empirically supported for interven- effect sizes were calculated using Cohen’s d statistic on tion with fluoxetine.11,12 The treatment manual included the primary outcome measure (i.e., CAPS total score) at instructions for monitoring of adverse effects and compli- posttreatment and at 6-month follow-up.
ance with prescribed dosage, tracking of PTSD symptomchange, and ongoing assessment of mental status. Starting dosage was 10 mg/day of fluoxetine (or pill placeboequivalent). Dosage was increased in 10-mg increments per week to a maximum of 60 mg/day, or until symptom Baseline group differences. Participants in the 3 treat-
remission was achieved. Increases or decreases in dosage ment conditions did not differ significantly on any demo- were based on physician judgment of clinical response graphic variable or on any baseline measure of psycho- and presence/absence of dose-limiting side effects. Mean fluoxetine dose across clients/sessions was 30 mg; 19 of Treatment dropout. Twelve people dropped out during
26 fluoxetine-condition completers received 30 mg or the 8-week treatment phase, leaving 76 treatment com- greater for the last 4 weeks of treatment. The modal fluox- pleters. There were no significant differences in dropout etine dose at end of treatment was 40 mg. The drug was rates on any baseline measure of psychopathology or across treatment conditions. Completers in each group Adherence to the pharmacotherapy protocol and dosing were 24 of 29 (83%) for EMDR, 26 of 30 (87%) for flu- schedule was monitored through weekly case review and oxetine, and 26 of 29 (90%) for pill placebo. Dropouts supervision of study pharmacotherapists.
were significantly younger than completers (dropoutmean age = 27.1 years, completer mean age = 37.6 years; F = 6.785, df = 1,86; p < .05). In addition, dropouts were Means and standard deviations for all primary and sec- more likely to have had child- versus adult-onset trauma ondary outcome measures are reported in Table 3. At post- treatment, the drop in total CAPS score was 59.0% for Clinician effects. Potential clinician-specific effects
EMDR, 46.0% for fluoxetine, and 43.6% for placebo, as were examined in 3 one-way ANCOVAs on posttreatment compared with 1-week pretreatment baseline assessment.
CAPS total score for the ITT sample, controlling for base- At 6-month follow-up, the CAPS total score drop was line symptom severity. Chi-square analyses were used to 62.2% for EMDR and 48.3% for fluoxetine.
compare dropout rates by clinician. Six clinicians had Notably, omnibus ANCOVA analyses (ITT and compl- each treated 5 or fewer participants and were combined eter) of the 3-condition model failed to reveal a significant within treatment condition for comparison with the re- effect of treatment on the primary outcome measures.
maining clinicians. No significant differences were found While common practice is to refrain from further analyses by clinician in posttreatment scores or dropout rates.
when the omnibus analyses reveal negative findings, be- Concurrent supportive psychotherapy. Potential im-
cause our study hypotheses emphasize 2-group outcomes pact of concurrent supportive psychotherapy at the time over omnibus outcomes, and in light of the absence of an of randomization was examined in the same manner for appropriate control group for the psychotherapy condi- both treatment completers and ITT sample. No significant tion, pairwise comparisons of the 3 conditions were con- main or interactive effects of concurrent treatment were ducted and their results reported below.
EMDR versus placebo. The posttreatment ITT (LOCF)
Treatment expectations. No group differences were
analyses revealed nonsignificant trends for greater remis- found on baseline treatment expectations for participants sion of PTSD in individuals receiving EMDR versus pla- randomly assigned to medication condition versus psy- cebo. In the completer analyses, EMDR was significantly chotherapy condition, nor were any significant interac- superior to placebo on reduction of PTSD symptoms and tions found between treatment group and expectations on showed a greater percentage of loss of diagnostic status study outcome. A positive main effect of baseline treat- ment expectations on PTSD symptom reduction at post- Fluoxetine versus placebo. Neither posttreatment ITT
treatment was observed (F = 5.8, df = 1,84; p = .018).
nor completer analyses revealed significant differences Postintervention treatment seeking. After cessation of
between fluoxetine and placebo on any outcome measure the study intervention, 4 participants (2 EMDR, 2 fluoxe- (Table 3). The majority of both fluoxetine and placebo tine) initiated nonstudy treatments. To assess the potential participants demonstrated loss of PTSD diagnosis; impact on 6-month follow-up findings, a second set of however, few participants achieved full remission of analyses was run in which data from these participants were extracted (completer analyses) or carried forward EMDR versus fluoxetine. As hypothesized, at post-
from last data point preceding onset of new treatment treatment, the 2 active treatments did not differ on mea- (ITF analyses). Analyses revealed no changes in direction sures of PTSD (Table 3). At 6-month follow-up, EMDR or significance level of findings. Thus, reported findings was superior to fluoxetine on sustained reduction of post- retain data from these participants.
traumatic symptoms (Table 3) for both ITF (LOCF) and Clinician treatment adherence. Videotapes of 24
completer analyses. EMDR was superior to fluoxetine in EMDR sessions (over 10% of 210 total study sessions) attaining complete remission of symptoms at 6 months were randomly selected for independent fidelity rating, posttreatment: 58% of EMDR subjects were asympto- with oversampling to ensure distribution across clinicians matic, compared with none in the fluoxetine group.
and session type. Fifty components across the 8 phases EMDR was also superior to fluoxetine in reduction of of treatment were included in the fidelity manual devel- self-reported depressive symptoms for both the ITF and oped for this protocol. The evaluator was an experienced, independent, certified EMDR clinician who was not per-sonally known to the study investigators. The evaluator reviewed videotapes and rated adherence according to a Impact of Trauma Onset on Treatment Outcome 4-point Likert-type scale: 0 = no adherence, 1 = some In order to account for the potential impact of index adherence, 2 = adherence acceptable, and 3 = adherence trauma onset on treatment outcome, ANCOVA analyses very good. The mean fidelity score across sessions was were rerun for continuous outcomes with a dummy-coded 2.57 (SD = 0.35; minimum = 1.76, maximum = 3.00).
onset variable. Main and interactive effects for onset and Table 3. Posttreatment and Follow-up Analyses for Primary and Secondary Outcome Measures by Sample and
Treatment Condition
< .001*
< .001*
< .001*
aNumbers represent p values for pairwise comparisons of continuous and categorical measures.
bResults at trend level or greater are indicated by an asterisk (*); p values significant at ≤ .05 are indicated with boldface.
cTotal of all 3 treatments combined.
Omnibus analysis of continuous measures used analysis of covariance with baseline as covariate; Pearson χ2 statistic was used for eAt posttreatment, p value is for 3-group comparison; at follow-up, p value is for 2-group (active treatment) comparison.
fDefined as CAPS total score below 20.
gAll active treatment completers are included in intent-to-follow analyses.
Abbreviations: BDI-II = Beck Depression Inventory-II, CAPS = Clinician-Administered PTSD Scale, EMDR = eye movement desensitization and reprocessing, NA = not applicable, PTSD = posttraumatic stress disorder.
treatment condition were entered into the model. “Child addition of the index trauma onset variable. No onset-by- onset” was defined as onset of index trauma prior to age treatment-condition interaction effects were observed at 18; “adult onset” was defined as index trauma onset at any time point. Onset was found to make additional con- or after age 18. Diagnosis and asymptomatic end-state tributions to outcome, as follows: immediately posttreat- function were examined by onset and treatment condi- ment, across treatment conditions, patients with adult- tion using χ2 analysis. Descriptives by treatment con- onset trauma showed significantly greater reduction in dition and trauma onset are reported in Table 4.
PTSD symptoms than those with child-onset trauma, for Baseline. Equivalent numbers of patients with child-
both ITT (p < .005) and completer (p = .02) samples.
onset (N = 45) and adult-onset (N = 43) index traumas These effects were maintained at 6-month follow-up (ITT, were randomly assigned to treatment. Treatment con- ditions did not differ in distribution of patients by onset The EMDR and placebo groups demonstrated larger (child onset: EMDR, N = 15 [51.7%]; fluoxetine, N = 13 effect sizes for adult- than for child-onset trauma, whereas [43.3%]; placebo, N = 17 [58.6%]). At baseline, patients fluoxetine exhibited the opposite pattern (Table 5).
with child-onset trauma demonstrated significantly PTSD diagnosis and asymptomatic end-state func-
higher PTSD symptoms on 1-month CAPS than patients tion. At posttreatment, differential effects were found by
with adult-onset trauma (child onset: mean = 77.71 onset of the index trauma (Table 4). Chi-square analyses [SD = 13.04]; adult onset: mean = 70.26 [SD = 12.87]; revealed that adult-onset patients were significantly more F = 7.281, df = 1,86; p < .01); this dropped to a trend for likely to both lose PTSD diagnosis (ITT, p = .052) and 1-week CAPS (child onset: mean = 73.49 [SD = 12.99]; achieve asymptomatic end-state function (ITT, p = .037) adult onset: mean = 68.74 [SD = 12.74]; F = 2.989, than child-onset patients (Figure 1). When probed by treatment condition, these differential effects were found CAPS total score. For both posttreatment and follow-
to occur only within the EMDR condition.
up analyses, main effects for treatment as reported above At 6-month follow-up, χ2 analyses revealed that within were retained at similar levels of significance with the the EMDR group, adult-onset patients were more likely to Table 4. Posttreatment and Follow-Up Completer Descriptives by Treatment Condition and
Trauma Onset
aTotal of all 3 treatments combined.
bDefined as CAPS total score below 20.
Abbreviations: CAPS = Clinician-Administered PTSD Scale, EMDR = eye movement desensitization and reprocessing, NA = not applicable, PTSD = posttraumatic stress disorder.
Table 5. Effect Sizes for Primary Outcome Measure (CAPS
Figure 1. Asymptomatic End-State Function (CAPS
total score) by Sample
score < 20) by Treatment Type and Index Trauma Onset
aAt 6-month follow-up, none of the fluoxetine participants achieved aPositive between-group effect sizes favor first treatment condition Abbreviations: CAPS = Clinician-Administered PTSD Scale, over second; negative between-group effect sizes favor second EMDR = eye movement desensitization and reprocessing, PTSD = posttraumatic stress disorder.
Abbreviations: CAPS = Clinician-Administered PTSD Scale, EMDR = eye movement desensitization and reprocessing, NA = notapplicable, PTSD = posttraumatic stress disorder.
achieve loss of PTSD diagnosis and achieve asympto- In this 8-week study that compared an exposure-based matic end state (ITF, p = .045) than those with childhood- psychotherapeutic treatment of PTSD (EMDR) with a onset trauma. None of the fluoxetine participants pharmacologic treatment (fluoxetine) and pill placebo, achieved asymptomatic end-state function, regardless of participants in all 3 groups sustained considerable im- age of trauma (Figure 1). EMDR patients sustained high provement. Eighty-eight percent of EMDR, 81% of fluox- rates of loss of diagnosis at 6-month follow-up, regardless etine, and 65% of placebo completers lost their PTSD of age of trauma, while fluoxetine adult-onset patients diagnosis, and 29% of EMDR, 15% of fluoxetine, and were significantly more likely to lose PTSD diagnosis 12% of placebo completers became asymptomatic (CAPS than fluoxetine child-onset patients (ITF, p = .036).
scores under 20). Over the 6 months following the cessa- tion of treatment, the EMDR group continued to mildly The clinical improvement in the EMDR group at 6- improve, while the fluoxetine group lost some of its gains: month follow-up was not confined to PTSD symptoms: at follow-up, 57% of EMDR completers were asympto- EMDR completers had significantly lower BDI-II scores matic, compared with none of the fluoxetine group.
than fluoxetine completers. Once the trauma is resolved, In this study, every treatment group improved substan- other domains of psychological functioning appear to im- tially: undergoing routine study procedures on a weekly prove spontaneously. This may have significant implica- basis seemed to have beneficial effects on most partici- tions for the treatment of depression in individuals with pants, regardless of treatment condition. A positive re- sponse to study participation is common in PTSD drug The issue of trauma onset has not been addressed in treatment studies: participating in a study generally ac- previous treatment outcome studies. The present study counts for more of the variance than the particular treat- found that participants in the EMDR condition with ment received (e.g., see references 12, 14, 15). Post- adult-onset index traumas had a substantially better treat- traumatic stress disorder has been shown to be quite ment response than those with childhood-onset trauma.
responsive to pill placebo on the order of a 35% to 40% Specifically, 100% of adult-onset participants lost di- improvement over baseline.45 There has been considerable agnostic status at posttreatment, versus a significantly debate whether thorough weekly assessments themselves lower 75% of child-onset participants. This distinction entail sufficient exposure therapy to make a clinically became even more pronounced at 6-month follow-up.
significant difference.45–47 The magnitude of the placebo Here, trauma onset significantly predicted a large dis- response in PTSD suggests that claims of treatment effi- tinction in end-state functioning: 75.0% of adult-onset cacy of any particular method need to be based on com- versus 33.3% of child-onset trauma participants receiving parisons with a placebo or inert-treatment group rather EMDR were asymptomatic at 6-month follow-up (see Figure 1). These findings support the efficacy of short- The effect size of the pill-placebo group in this study term treatment with EMDR in the resolution of traumatic was larger than that of any other placebo groups in the sequelae associated with adult-onset trauma. In contrast, PTSD treatment outcome literature—in fact, larger than for most individuals with childhood-onset trauma (all of that of several active treatments that claim efficacy com- whom, in this study, were victims of chronic intrafamilial pared with wait-list control groups. In order to study physical and/or sexual abuse), 8 weeks of therapy was changes in psychophysiologic reactivity among the dif- not enough to resolve longstanding trauma imprints and ferent treatments, all subjects in this study were exposed to 2 personalized trauma scripts that involved intense con- In contrast to the EMDR condition, the outcome for frontation with their traumatic memories. This exposure participants in the fluoxetine condition did not vary as may have played a significant role in the positive outcome a function of trauma onset. Whereas participants in across treatment conditions. Exposure to memories of the fluoxetine condition exhibited a robust decrease one’s trauma has repeatedly been identified as a critical in PTSD symptomatology, significantly fewer were en- element in effective PTSD therapy.48,49 Exposure may also tirely asymptomatic at posttreatment, and none remained have contributed to the large effect size in the fluoxetine asymptomatic at 6-month follow-up. This distinction is group, which was larger than those previously reported in important, since recent research has revealed that patients with subthreshold PTSD have similar degrees of impair- The effect size of the EMDR group was comparable to ment in social and work functioning as those with full- that of previous EMDR studies (e.g., see reference 50).
blown PTSD.52,53 This suggests that brief treatment with The EMDR responders not only maintained their treat- an SSRI alone is insufficient to resolve posttraumatic ment gains after the study but continued to improve psychopathology for most patients with PTSD.
slightly over time: at 6-month follow-up, 57% of EMDR At 6-month follow-up, fluoxetine treatment produced completers had a CAPS score below 20 (asymptomatic a significantly larger effect size than EMDR for the sub- end-state function), compared with none of the fluoxetine group of individuals with child-onset trauma. This was group. It is not surprising that fluoxetine subjects lost primarily due to the smaller standard deviation within the some of their improvement at follow-up—pharmacologic fluoxetine condition. Whereas EMDR participants with effects cannot be expected to last over time, but the fact child-onset trauma varied markedly in PTSD symptom- that the effects of EMDR were maintained, and somewhat atology at follow-up (e.g., some were completely asymp- improved, has important clinical and economic implica- tomatic while others showed little progress or slight tions. Continued improvement after treatment cessation worsening of symptoms), child-onset PTSD participants has previously been reported with prolonged exposure8 who received fluoxetine experienced a more consistently and EMDR.51 This suggests that, once people deal with modestly positive response to pharmacotherapy than to their traumatic memories, they are likely to continue to EMDR. This suggests that, while brief intervention with improve without further intervention.
fluoxetine is unlikely to resolve PTSD for adults with child-onset trauma, it provides moderate symptom reduc- tion, with little risk of symptom exacerbation.
1. Kessler RC, Sonnega A, Bromet E, et al. Posttraumatic stress disorder In this study, fluoxetine did not do better than placebo, in the National Comorbidity Survey. Arch Gen Psychiatry 1995;52: not because participants receiving fluoxetine did not im- prove but because the placebo response in this study was 2. Tjaden P, Thoennes N. Full Report of the Prevalence, Incidence and Consequences of Violence Against Women: Findings From the National so robust. Recent studies have shown that patients who Violence Against Women Survey. Washington, DC: National Center of receive prolonged treatment with an SSRI (beyond the Justice; November 2000; Publication NCJ 183781 customary 8- to 12-week studies) are likely to experience 3. Breslau N, Davis GC, Andreski P, et al. Traumatic events and post- traumatic stress disorder in an urban population of young adults.
a gradual decrease in PTSD symptomatology.54,55 In one study,55 32 weeks of treatment with sertraline achieved an 4. Kessler RC, Zhao S, Katz SJ, et al. Past-year use of outpatient services end state comparable to that seen in the EMDR condition for psychiatric problems in the National Comorbidity Survey. Am JPsychiatry 1999;156:115–123 in the present study. However, that long-term study did 5. Kilpatrick DG, Saunders BE. Prevalence and Consequences of Child not address the issue of whether improvement was sus- Victimization: Results from the National Survey of Adolescents: Final tained after cessation of the medication.
Report. Washington, DC: US Department of Justice, Office of JusticePrograms; 1997 Clearly, more data are needed to determine which 6. Hoge CW, Auchterlonie JL, Milliken CS. Mental health problems, use of patients with chronic childhood trauma and/or multiple mental health services, and attrition from military service after returning comorbidities will respond best to trauma processing/ from deployment to Iraq or Afghanistan. JAMA 2006;295:1023–1032 7. Foa EB, Dancu CV, Hembree EA, et al. A comparison of exposure exposure-based therapies versus long-term pharmaco- therapy, stress inoculation training, and their combination for reducing logic treatment. Studies are also needed to establish posttraumatic stress disorder in female assault victims. J Consult Clin whether there is a synergistic effect between pharmaco- 8. Resick PA, Schnicke MK. Cognitive processing therapy for sexual therapy and exposure therapy for PTSD, a combination assault victims. J Consult Clin Psychol 1992;60:748–756 that is common in clinical practice, but whose efficacy 9. Carlson JG, Chemtob CM, Rusnak K, et al. Eye movement desensitiza- tion and reprocessing for combat-related posttraumatic stress disorder.
10. Wilson S, Becker LA, Tinker RH. Eye movement desensitization and reprocessing (EMDR): treatment for psychologically traumatized indi-viduals. J Consult Clin Psychol 1995;63:928–937 11. van der Kolk BA, Dreyfuss D, Michaels M, et al. Fluoxetine in Patients with trauma histories, PTSD, and depression posttraumatic stress disorder. J Clin Psychiatry 1994;55:517–522 are ubiquitous in clinical practice. This study demon- 12. Connor KM, Sutherland SM, Tupler LA, et al. Fluoxetine in post- strates that the vast majority of patients with adult-onset traumatic stress disorder: randomized, double-blind study. Br JPsychiatry 1999;175:17–22 PTSD can recover with a short period of intense, 13. Brady K, Pearlstein T, Asnis GM, et al. Efficacy and safety of sertraline exposure-based treatment, with lasting positive results.
treatment of posttraumatic stress disorder: a randomized controlled trial.
Merely paying careful attention to the patients’ symp- 14. Davidson JR, Rothbaum BO, van der Kolk BA, et al. Multicenter toms, as was done here in the placebo condition, as well double-blind comparison of sertraline and placebo in the treatment of as treating with SSRIs, can be helpful, particularly in pa- posttraumatic stress disorder. Arch Gen Psychiatry 2001;58:485–492 tients with childhood-onset trauma. However, these ap- 15. Marshall RD, Beebe KL, Oldham M, et al. Efficacy and safety of paroxe- tine treatment for chronic PTSD: a fixed-dose, placebo-controlled study.
proaches do not lead to complete symptom remission and the benefits do not endure with time. The present study 16. Klosko JS, Barlow DH, Tassinari R, et al. A comparison of alprazolam supports the empirical literature that proposes that skilled and behavior therapy in the treatment of panic disorder. J Consult ClinPsychol 1990;58:77–84 confrontation with traumatic memories within a safe 17. Black DW, Wesner R, Bowers W, et al. A comparison of fluvoxamine, therapeutic setting is the treatment of choice for PTSD cognitive therapy, and placebo in the treatment of panic disorder.
Arch Gen Psychiatry 1993;50:44–50 with adult-onset trauma. Future research should assess 18. Spiegel DA, Bruce TJ. Benzodiazepines and exposure-based cognitive the impact of lengthier interventions, combination treat- behavior therapies for panic disorder: conclusions from combined ments, and treatment sequencing on the resolution of treatment trials. Am J Psychiatry 1997;154:773–781 PTSD in adults with childhood-onset trauma.
19. Cox BJ, Swinson RP, Morrison B, et al. Clomipramine, fluoxetine and behavior therapy in the treatment of obsessive-compulsive disorder: a meta-analysis. J Behav Ther Exp Psychiatry 1993;24:149–153 Drug names: fluoxetine (Prozac and others), sertraline (Zoloft 20. Baxter LR Jr, Schwartz JM, Bergman KS, et al. Caudate glucose metabolic rate changes with both drug and behavior therapy forobsessive-compulsive disorder. Arch Gen Psychiatry 1992;49:681–689 Acknowledgments: The authors thank Jose Hidalgo, M.D.; Jeffrey 21. DeRubeis RJ, Gelfand LA, Tang TZ, et al. Medications versus cognitive Weir, L.I.C.S.W.; Deborah Rozelle, Ph.D.; Paula Morgan-Johnson, behavior therapy for severely depressed outpatients: mega-analysis of L.I.C.S.W.; Jelica Todosijevich, Ph.D.; Tony Luxenburg, Ph.D.; Caren four randomized comparisons. Am J Psychiatry 1999;156:1007–1013 Swift, Ph.D.; Miriam Kissin, Ph.D.; Susan Rogers, Ph.D.; Kate Walsh, 22. Elkin I, Shea MT, Watkins JT, et al. National Institute of Mental Health Ph.D.; Elizabeth Nickrenz, M.A.; Ibrahim Dager, M.D.; Ruth Lanius, Treatment of Depression Collaborative Research Program: general M.D.; Terry Smith, M.D.; Isabelle Soulard, M.D.; Dan Siskind, M.D.; effectiveness of treatments. Arch Gen Psychiatry 1989;46:971–982 Nancy Smythe, M.S.W., Ph.D.; and Richard LaDue for the administra- 23. Antonuccio DO, Danton WG, DeNelsky GY. Psychotherapy versus tion of treatment protocols and outcome instruments or for assistance medication for depression: challenging the conventional wisdom in the data analysis. These acknowledged individuals report no con- with data. Professional Psychol Res Pract 1995;26:574–585 24. Klein DF. Flawed meta-analyses comparing psychotherapy with pharmacotherapy. Am J Psychiatry 2000;157:1204–1211 40. Pitman RK, Orr SP, Forgue DF, et al. Psychophysiologic assessment 25. Nemeroff CB, Heim CM, Thase ME, et al. Differential responses to of posttraumatic stress disorder imagery in Vietnam combat veterans.
psychotherapy versus pharmacotherapy in patients with chronic forms of major depression and childhood trauma. Proc Natl Acad Sci U S A 41. Korn D, Spinazzola J. EMDR Treatment and Fidelity Manual.
Boston, Mass: The Trauma Center; 2004. Available at: 26. Van Etten ML, Taylor S. Comparative efficacy of treatments for post- http://www.traumacenter.org. Accessibility verified Nov 15, 2006 traumatic stress disorder: a meta-analysis. Clin Psychol Psychother 42. Shapiro F. Eye Movement Desensitization and Reprocessing: Basic Principles, Protocols, and Procedures. New York, NY: Guilford; 1995 27. Otto MW, Penava SJ, Pollack SJ, et al. Cognitive, behavioral and 43. Frison L, Pocock SJ. Repeated measures in clinical trials: analysis using pharmacologic perspectives on the treatment of post-traumatic stress mean summary statistics and its implications for design. Stat Med 1992; disorder. In: Pollack MH, Otto MW, Rosenbaum JF, eds. Challenges in Clinical Practice: Pharmacologic and Psychosocial Strategies.
44. Fleiss JL. Measures of effect size for categorical data. In: Cooper HM, New York, NY: Guilford Press; 1996:219–260 Hedges LV, eds. The Handbook of Research Synthesis. New York, NY: 28. US Department of Veterans Affairs and Department of Defense. VA/ DoD Clinical Practice Guideline for the Management of Post-Trau- 45. Krakow B, Hollifield M, Warner TD. Placebo effect in posttraumatic matic Stress. Washington, DC: US Department of Veterans Affairs stress disorders [letter]. JAMA 2000;284:563–564 and Department of Defense; 2004. Available at: http:// 46. Davidson JR, Malik ML, Sutherland SN. Response characteristics to www.oqp.med.va.gov/cpg/PTSD/PTSD_Base.htm. Accessibility antidepressants and placebo in post-traumatic stress disorder. Int Clin 29. Chambless DL, Baker MJ, Baucom DH, et al. Update on empirically 47. Baker DG, Diamond BI, Gillette G, et al. A double-blind, randomized, validated therapies, 2. Clin Psychol 1998;51:3–16 placebo-controlled, multi-center study of brofaromine in the treatment 30. American Psychiatric Association. Practice Guideline for the Treatment of post-traumatic stress disorder. Psychopharmacology (Berl) 1995;122: of Patients With Acute Stress Disorder and Posttraumatic Stress Disor- der. Am J Psychiatry 2004;161(suppl 11):1–31 48. Foa EB, Keane T, Friedman M. Effective Treatments for PTSD.
31. Brom D, Kleber RJ, Defares PB. Brief psychotherapy for posttraumatic stress disorders. J Consult Clin Psychol 1989;57:607–612 49. Foa EB, Rothbaum BO, Riggs DS, et al. Treatment of posttraumatic 32. American Psychiatric Association. Diagnostic and Statistical Manual stress disorder in rape victims: a comparison between cognitive- of Mental Disorders, Fourth Edition. Washington, DC: American behavioral procedures and counseling. J Consult Clin Psychol 1991; 33. Spinazzola J, Blaustein M, van der Kolk BA. Posttraumatic stress 50. Rothbaum BO. A controlled study of eye movement desensitization disorder treatment outcome research: the study of unrepresentative and reprocessing in the treatment of posttraumatic stress disordered samples? J Trauma Stress 2005;18:425–436 sexual assault victims. Bull Menninger Clin 1997;61:317–334 34. Blake DD, Weathers FW, Nagy LM, et al. The development of a 51. Wilson SA, Becker LA, Tinker RH. Fifteen-month follow-up of eye Clinician-Administrated PTSD Scale. J Trauma Stress 1995;8:75–90 movement desensitization and reprocessing (EMDR) treatment for 35. Weathers FW, Ruscio AM, Keane TM. Psychometric properties of posttraumatic stress disorder and psychological trauma. J Consult nine scoring rules for the Clinician-Administered Posttraumatic Stress Disorder Scale. Psychol Assess 1999;11:124–133 52. Zlotnick C, Franklin CL, Zimmerman M. Does “subthreshold” posttrau- 36. Weathers FW, Keane TM, Davidson JR. Clinician-Administered PTSD matic stress disorder have any clinical relevance? Compr Psychiatry Scale: a review of the first ten years of research. Depress Anxiety 2001; 53. Stein MB, Walker JR, Hazen AL, et al. Full and partial posttraumatic 37. First MB, Spitzer RL, Gibbon M, et al. Structured Clinical Interview stress disorder: findings from a community survey. Am J Psychiatry for DSM-IV Axis I Disorders (SCID-I). Washington, DC: American 54. Rapaport MH, Endicott J, Clary CM. Posttraumatic stress disorder and 38. First MB, Spitzer RL, Gibbon M, et al. Structured Clinical Interview quality of life: results across 64 weeks of sertraline treatment. J Clin for DSM-IV Axis II Disorders (SCID-II). Washington, DC: American 55. Londborg PD, Hegel MT, Goldstein S, et al. Sertraline treatment of 39. Beck AT, Steer RA. Manual for the Revised Beck Depression Inven- posttraumatic stress disorder: results of 24 weeks of open-label continu- tory. San Antonio, Tex: Psychological Corporation; 1993 ation treatment. J Clin Psychiatry 2001;62:325–331

Source: http://ww.traumacenter.org/products/pdf_files/van_der_Kolk_JCLINFINAL.pdf